Enabling robots to flexibly schedule and compose learned skills for novel long-horizon manipulation under diverse perturbations remains a core challenge. Early explorations with end-to-end VLA models show limited success, as these models struggle to generalize beyond the training distribution. Hierarchical approaches, where high-level planners generate subgoals for low-level policies, bring certain improvements but still suffer under complex perturbations, revealing limited capability in skill composition. However, existing benchmarks primarily emphasize task completion in long-horizon settings, offering little insight into compositional generalization, robustness, and the interplay between planning and execution. To systematically investigate these gaps, we propose RoboHiMan, a hierarchical evaluation paradigm for compositional generalization in long-horizon manipulation. RoboHiMan introduces HiMan-Bench, a benchmark of atomic and compositional tasks under diverse perturbations, supported by a multi-level training dataset for analyzing progressive data scaling, and proposes three evaluation paradigms (vanilla, decoupled, coupled) that probe the necessity of skill composition and reveal bottlenecks in hierarchical architectures. Experiments highlight clear capability gaps across representative models and architectures, pointing to directions for advancing models better suited to real-world long-horizon manipulation tasks. Videos and open-source code can be found on our project website: https://chenyt31.github.io/robo-himan.github.io/.

翻译：使机器人能够灵活调度和组合已学技能，以应对多样化扰动下的新颖长时程操作，仍然是一个核心挑战。早期基于端到端视觉语言动作（VLA）模型的探索成果有限，因为这些模型难以泛化至训练分布之外。采用高层规划器为低层策略生成子目标的分层方法带来了一定改进，但在复杂扰动下仍表现不佳，显示出技能组合能力的局限。然而，现有基准主要关注长时程场景下的任务完成度，对组合泛化、鲁棒性以及规划与执行间的交互关系揭示不足。为系统性地探究这些不足，我们提出了RoboHiMan——一个面向长时程操作组合泛化的分层评估范式。RoboHiMan引入了HiMan-Bench，这是一个包含多样化扰动下原子任务与组合任务的基准，并辅以用于分析渐进式数据缩放的多层级训练数据集；同时提出了三种评估范式（基础型、解耦型、耦合型），用以探究技能组合的必要性并揭示分层架构中的瓶颈。实验凸显了代表性模型与架构间显著的能力差距，为推进更适应现实世界长时程操作任务的模型指明了方向。视频与开源代码可在我们的项目网站查看：https://chenyt31.github.io/robo-himan.github.io/。